Design of information flow for networked control systems

Más información sobre el libro

Rapid advances in information processing, communication, and sensing technologies have enabled devices to incorporate embedded processors, networking capabilities, and sensors. This evolution allows for an architecture where numerous simple components communicate and cooperate to achieve a collective goal. Such a distributed architecture offers enhanced performance, reliability, and design simplicity. However, it necessitates an extension of traditional control, communication, and computation theories, advocating for a unified understanding of these fields. From an estimation and control standpoint, real communication channels can introduce non-classical information patterns, leading to significant performance losses. This text addresses new design principles to mitigate such degradation, focusing on the joint design of information flow and control laws. Unlike traditional control design, which assumes a specific information flow to calculate optimal control inputs, this approach synthesizes optimal information flow alongside control laws that meet the constraints of that flow. Consequently, it prompts critical questions: What actions should an agent take? Who should they communicate with? What information should be shared? When should communication occur? This joint design of information flow and control input represents a crucial degree of freedom for system designers, yielding substantial performance improvements over conve